Repository navigation
Backend not-empty rule disagrees with Arc .NET on which characters are whitespace #188
Description
Problem
Screenplay decision 0071 (https://github.com/Cratis/Screenplay/blob/main/decisions/0071-whitespace-only-text-does-not-satisfy-not-empty.md) fixes not empty on text to the Unicode White_Space set, the same set .NET char.IsWhiteSpace and FluentValidation's NotEmpty() use. Arc.TypeScript's backend rule uses JavaScript trim() (Source/Core/validation/evaluateRule.ts, empty), which treats U+FEFF as whitespace and U+0085 as a character, so a whitespace-only value is accepted or rejected differently than by Arc .NET and the Arc frontend (Cratis/Arc#3224).
Requested
Treat a string as empty when it consists only of Unicode White_Space characters (for example /^\p{White_Space}*$/u), and add specs for U+0085, U+FEFF and an ordinary whitespace-only value.
Status
State: Done — Delivered by merged PR #194 at 862803d; issue closed completed. Full review and PR CI passed.
Branch: (#194)
Scope: Unicode White_Space for empty/notEmpty string validation; one patch PR.
Plan: Replace the trim()-based string emptiness predicate in Source/Core/validation/evaluateRule.ts with /^\p{White_Space}*$/u for both empty and notEmpty; add for_evaluateRule specs for U+0085, U+FEFF, ordinary whitespace, empty string and mixed text.
Delivered: Nothing yet.
Remaining: To be confirmed.
Follow-ups: None.
Activity
woksin commentedon Oct 29, 2026
Analysis and proposed solution
Current behaviour
Source/Core/validation/evaluateRule.ts:8treats a string as empty when JavaScripttrim()removes its entire content. ThenotEmptyandemptyrules both use that predicate atSource/Core/validation/evaluateRule.ts:27-28. Consequently U+0085 is treated as content, whereas U+FEFF is treated as whitespace.Root cause / gap
ECMAScript trimming and Unicode
White_Spaceare different character sets. The shared predicate implements JavaScript trimming semantics rather than the cross-stack text-validation contract described in this issue. Changing only thenotEmptybranch would also leave the complementaryemptyrule inconsistent.Proposed change
Replace the string-specific trim check in
Source/Core/validation/evaluateRule.tswith an anchored Unicode-property expression such as/^\p{White_Space}*$/u. Leave null, number, boolean, array and Guid handling unchanged.Add focused specs under
Source/Core/validation/for_evaluateRule/when_notEmptyandwhen_empty, following thedescribe, asynchronousbeforeEach,itandshouldassertion style demonstrated bySource/Core/validation/for_evaluateRule/when_notEmpty/with_guid_empty.ts:7-15. Verify that U+0085 and ordinary whitespace-only strings failnotEmpty, U+FEFF passes it, andemptyreturns the inverse. Include empty strings and text mixed with whitespace as controls; preserve the existing non-string regression cases.Size, risk and release intent
S; low implementation risk, with an intentional validation-behaviour correction for affected characters; patch.
Dependencies and blockers
Coordinate parity with Cratis/Arc#3224. No backend implementation blocker is evident; the Unicode contract is specified in the linked Screenplay decision.
woksin commentedon Oct 29, 2026
Delivered in #194: empty and notEmpty now use Unicode White_Space, including regression coverage for U+0085 and U+FEFF.